Grocery Shopping for America Terrorism, National Identity, and Consumer Behavior

Transcription

1 Grocery Shopping for America Terrorism, National Identity, and Consumer Behavior Adam G. Hughes, University of Virginia Sonal Pandya, University of Virginia Rajkumar Venkatesan, University of Virginia November 14, 2014

2 Does national identity have a causal effect on economic behavior?

3 Does national identity have a causal effect on economic behavior?

4 National Identity and Political Behavior: Internal vs. External Validity Survey experiments provide strong internal validity, but... Behavior consistent with survey responses? Does salience of national identity vary over time/across context?

5 Universal, Consistent, and Frequent Behavior: Supermarket Shopping Average US household shops once a week Heavier reliance on heuristics for low cost purchases Data allow us to control for changes in price, availability, and previous consumption

6 Nationality-Based Marketing Strategies Link National Identity to Behavior

7 Terrorism: Exogenous Shock Makes National Identity More Salient

8 Terrorist Events Increases Market Share of American-Sounding Brands Identifying assumption: events only influence purchasing choice via change in association with American identity Controls for price, availability, seasonality

9 Behavior: Weekly Supermarket Purchases Nationally representative sample from IRI (academic-use data set) The scanner data includes: 5184 brands (i), 1154 supermarkets (j), 22 product categories (k), 156 weeks (t) - all weeks in 2001, 2002, 2003

10 50 Geographic Markets

11 Outcome: Weekly Change in Market Share Growth Rate Share01 ijkt is the number of units of the brand sold, as a percent of all units in the product category sold for brand i-product category k in store j in week t First difference controls for all time-invariant characteristics of brands, including supply and demand

12 AmericanScore i = N Coders who Perceive Brand as American

13 AmericanScore i = 7

14 AmericanScore i = 0

15 Empirical Model: Supermarket Sales in Week After 9/11 Share01 ijkt = β 1 T errorattack+ β 2 AmericanScore i + β 3 T errort hreat t AmericanScore i + β 4 Price01 ijkt 1 + β 5 NumVariants01 ijkt 1 + ɛ ijkt 1

16 Predicted Effect: 9/11 Increases American-Sounding Brand Market Share

17 Largest Share Increases for Most American-Sounding Brands 2001: : all brands avg. change : AmericanScore i = 5 avg. change : AmericanScore i = 7 avg. change AmericanScore i = 7 brands: 29% increase in market share relative to previous week. By comparison, average price increase generates 4% increase.

18 DHS Terror Threat Level: Threat Without Supply-Side Effects 9/11 attack: confounding unobserved effects? 2003 DHS elevated terror alert: No confounding effects Uniform effect across the US Year over year market share changes controls for all time-invariant characteristics

19 Predicted Effect: Threat Level Increases American-Sounding Brand Market Share

20 Who Switched into American-Sounding Brands? Store customers: population for 2-mile radius around each store Correlates of stronger ex ante attachment to US identity: Members of the armed forces (9/11) US citizens (DHS) Blue collar workers (DHS)

21 Survey Experiment: Importance of Nationalism and Persistence of Shock? Experimental Design Recruit 276 subjects from MTurk in 2014 Randomize national brand cues (American/non-American) Randomize perceived threat (9/11 memory task/y2k memory task) Brand preference questions

22 Invented American and Non-American Brand Pairs Control Treatment

23 Nationalistic Subjects Prefer American-Sounding Brands Predicted Willingness to Purchase Predicted American Product Evaluation Nationalism Nationalism

24 Nationalism Correlates Most with Brand Choice Brand preferences uncorrelated with item price, respondent education, income, and partisanship. No effect of 9/11 treatment in 2014; no long-term threat persistence.

25 Conclusion: Real World Behavior Matters Supermarket purchases: real-time observable responses to exogenous shocks across contexts Novel measure: supermarket purchases link identity and behavior via branding IR research: unique behavioral correlate of international policy preferences